Journal of Electronic Science and Technology, Volume. 22, Issue 2, 100250(2024)

Data augmentation method for insulators based on Cycle-GAN

Run Ye1...3,*, Azzedine Boukerche2, Xiao-Song Yu1, Cheng Zhang3, Bin Yan1,3, and Xiao-Jia Zhou13 |Show fewer author(s)
Author Affiliations
  • 1School of Automation Engineering, University of Electronic Science and Technology of China, Chengdu, 611731, China
  • 2School of Electrical Engineering and Computer Science, University of Ottawa, Ottawa, K1N6N5, Canada
  • 3Yangtze River Delta Research Institute (Huzhou), University of Electronic Science and Technology of China, Huzhou, 313001, China
  • show less
    Figures & Tables(13)
    Structure of attention Cycle-GAN.
    Attention module structure diagram.
    Channel attention network diagram.
    Self-attention network diagram.
    Transfer training GAN structure diagram
    Example of converting glass insulators into composite insulators.
    Example of converting composite insulators into glass insulators.
    Example of converting ceramic insulators into glass insulators
    Example of converting glass insulators into ceramic insulators
    Example of defect sample generation.
    • Table 1. List of experimental equipment.

      View table
      View in Article

      Table 1. List of experimental equipment.

      Device nameModel numberQuantity
      CPUi7-12700k1
      GPURTX3080 12G1
      MainboardROG STRIX Z690-E1
      RAMDDR5 16G2
      Hard diskm2 Solid state drive 1T1
    • Table 2. Comparison of sample conversion indicators of Cycle-GAN insulators based on the attention mechanism.

      View table
      View in Article

      Table 2. Comparison of sample conversion indicators of Cycle-GAN insulators based on the attention mechanism.

      indexQuestCycle-GANDistance GANOurs
      PSNRGlass→Ceramic18.22412.13924.543
      Ceramic→Glass18.19011.94023.939
      Glass→Composite17.935211.94023.446
      Composite→Glass17.99012.726223.105
      SSIMGlass→Ceramic0.6870.2630.938
      Ceramic→Glass0.7030.2910.923
      Glass→Composite0.6830.2780.921
      Composite→Glass0.6900.2800.913
    • Table 3. Comparison of defect sample indicators based on background PSNR, SSIM, and FID.

      View table
      View in Article

      Table 3. Comparison of defect sample indicators based on background PSNR, SSIM, and FID.

      IndexDCGANWGAN-GPOurs
      FID25.3318.7812.54
      PSNR15.4619.1223.76
      SSIM0.580.670.92
    Tools

    Get Citation

    Copy Citation Text

    Run Ye, Azzedine Boukerche, Xiao-Song Yu, Cheng Zhang, Bin Yan, Xiao-Jia Zhou. Data augmentation method for insulators based on Cycle-GAN[J]. Journal of Electronic Science and Technology, 2024, 22(2): 100250

    Download Citation

    EndNote(RIS)BibTexPlain Text
    Save article for my favorites
    Paper Information

    Category:

    Received: Jul. 13, 2023

    Accepted: Apr. 9, 2024

    Published Online: Aug. 8, 2024

    The Author Email: Ye Run (rye@uestc.edu.cn)

    DOI:10.1016/j.jnlest.2024.100250

    Topics